WO 2009/026258 PCT/US2008/073526 Feature Dependent Extended Depth of Focusing on Semi Transparent Biological Specimens RELATED APPLICATIONS This application claims the benefit of U S. Application Set No 12/193,666 filed on August 18, 2008 which claims the benefit oflpiovisional apple. 60/956,504 filed on August 17, 2007 FIELD OF THE INVENTION The present invention relates to the field of digital processing of microscope images, specifically to an extended depth-of-focus (EDF) of thick or three-dimensional semi-transparent biological specimens. BACKGROUND OF THE INVENTION High-resolution microscopy suffers forn limited depth of field, which prevents thick or uneven specimen preparations from being imaged entirely in fbcus. Objects that appear outside the narrow depth of field or focal plane become quickly blurred and out of focus, forcing the microscopist to constantly manually focus back and fbrth. This not only limits the productivity of the microscopist but also increases the likelihood he or she will miss a subtle feature that may appear only in a narrow focal plane. Moreover, this limited depth of field becomes worse as the magnification increases because it is directly dependent on the numerical aperture of the microscope objective, which incr eases with the magnification. This prevents a microscopist from using the highest magnification objective on a consistent basis, and forces him/he to strike a balance between magnification and the labor-intensive task of constant up and down focusing.
WO 2009/026258 PCT/US2008/073526 Thick tissues, that are thicker than the depth of field of a microscope objective, introduces a third dimension to a biopsy sample, in effect making the specimen, or portions thereof, three-dimensional in space, A consequence of the three-dimensional character of a specimen is that the cellular material is located at various focal planes, which thereby requires constant focusing and refocusing to observe cells at various contours of the sample Specifically, when obtaining a non-lacerational bush biopsy of a tissue, a brush is used which is sufficiently stiff so as to penetrate the various layers of epithelium. In the process of obtaining a full thickness tissue specimen, tissue fragments in addition to single cells and cell cluster s ar e obtained and transferred onto a microscope slide. This occurs, when the brush biopsy instrument disclosed in US Patent No 6,258,044, incorporated herein by reference is used to sample otal epithelial tissue Similarly, when using the brush biopsy instruments for sampling tissue of the gastrointestinal tract disclosed in US Patent No's 6,494,845; 6,676,609 and 7,004,913, all of which are incorporated herein by reference, the resultant specimen contains single cells, cell clusters and thick tissue fragments. These specimens are markedly different fiom the cell monolayers prepared foi the analysis of exfoliative cytological specimens, whereby only a superficial sweep of a tissue is conducted, and rio tissue fragments are obtained This novel specimen, containing single cells, cell clusters and tissue fragments, is essentially a hybrid between a cytological smear, and histological sections. The ability to view tissue fi agments, in addition to single cells, confer s an enormous informational advantage to a pathologist in making a diagnosis Intact tissue provides the pathologist with important information about a tissue's architecture, which is not available in cytological smears This benefit is especially critical in the evaluation of gastrointestinal tissue, which is a complex tissue containing various cell types including glandular and columnar epithelium. Furthermore, U.S. patent No 's 6,297,044 and 6,284,482 - both of which are incorporated herein by reference - disclose a computerized system fbi analyzing and 2 WO 2009/026258 PCT/US2008/073526 classifying these novel specimens consisting of disaggregated cells and tissue fragments. In conducting its analysis of a slide, the computer scores and classifies cells that are most abnormal within a population of cells based on morphological criteria. As such, when analyzing a specimen that has dimensionality, often there will be areas of the specimen that are out of focus. As a result, the computer may classify cells found in these out of focus ar eas as abnormal because they may appear to the computer as exhibiting features that it is trained to classify as abnormal Composing a two-dimensional image out of the three-dimensional specimen would combine the respective advantages of each A pathologist would be able to capture the information available from a three-dimensional sample without the drawbacks associated with the constant need to focus and refocus the microscope. This would additionally, make the computer analysis of such samples more sensitive, as normal cells that are out of focus and thusly appear to the computer as abnormal, would be eliminated or significantly reduced This can be achieved through extended depth of focus (EDF) processing techniques, which are well known in the prior art See, for example, U.S. Patent No 4,584,704 With EDF processing an automated microscope captures a set of images taken at differ ent fbcal planes at the same location on a slide and then combines these images into a composite image A single image appears sharp and well-focused only in those areas where e the focal plane intersects the three-dimensional contours of the sample A single high-resolution image cannot be in-focus everywhere With EDF processing an automated microscope captures a set of images taken at regular z-intervals at the same location and then recovers from each slice those pixels that are in focus to build a single composite images fiom the in focus pixels. Once the in-focus pixels of each image are fused into one image, the resultant image is essentially in-focus everywhere However, the EDF algorithms of the prior art - which are ideal for creating a composite image of the top suiftce of a three-dimensional object - do not work well for biological specimens, which present an additional complexity of having multiple semi 3 WO 2009/026258 PCT/US2008/073526 transparent objects or cells stacked on each other. This is because in composing a composite image, standard EDF algorithms blindly extract the sharpest pixels from each focal plane, raising the possibility that a composite cell image contains pixels coming fiom multiple cells that happened to be situated on top of each other. In that instance a composite image may appear to represent a single cell, when in fact there were several cells stacked on top of each other, each of which could be observed by a microscope due to their semi-transparent character. For example, in U.S. Patent No. 4,661,986 the best focused pixels are selected to be incorporated into the composite image In U S. Patent No 7,058,233 a composite image is constructed by selecting well-focused edges or boundaries. These methods are of no utility when attempting to construct a composite image of a semi-ti ansparent biological sample for diagnostic purposes, whereby it is critical to have all the pixels come from a single object or cell and to avoid pixel contamination In order for an EDF algorithm to be properly applied to thick, semi transparent biological specimens it must take into account which pixels belong to which object, and preserve most of the pixels of'that cell or object even though they may not necessarily be sharp. Additionally, for diagnostic purposes, a pathologist prefers to see a composite image that only includes cell nuclei. This presents a further limitation of standard EDF algorithms; the nucleus' sharpness might be less than the cell boundary's sharpness, causing the composite image to only display the cell boundary and not the diagnostic important cell nucleus. This problem is particularly pronounced in the case of columnar epithelial cells. Columnar cells, which are found in the lining of the gastro-intestinal tract, are characteristically tall or oblong with its nucleus usually situated in its middle to lower region. When viewing a sample of columnar cells that are vertically oriented on a slide, such that the microscope is looking down the axis of a cylindrical cell, the cell boundaries may be in best focus at the top of the sample while the nucleus is in-focus at the bottom A pathologist needs to see the nucleus, but prior art EDF algorithms, which select in focus images based solely on the sharpness of the image, cannot distinguish between 4 WO 2009/026258 PCT/US2008/073526 different objects, both of which appear sharp and in-focus As a result, both the cell boundary and the cell nucleus will be selected, or in some instances only the cell boundary will be selected. In the latter situation, the resultant composite image will feature an array of abutting cell boundaries, and will take on the appearance of fish scales. This image will be sharp and in-focus everywhere, but will not convey any diagnostically important information to an examining pathologist or to a specialized computer system that may be analyzing the slide. In summary, the semi-transpar ent quality of biological specimens pr esents a problem with applying prior art EDF algorithms to thick cellular specimens. The prior algorithms were designed to create a composite image of the outside surface of a three dimensional object With biological specimens, on the other hand, the area of interest is not limited to the outside surface, but rather there are areas of interest below the outside surface of a specimen Due to light absorption within a tissue, structures on the surfhee have a tendency to have a higher contrast or sharpness than structures inside a transparent tissue. Because tissue is comprised of a mass of cells, a pathologist needs to see the cells and nuclei that are situated below the outside surface, which comprise the thickness of the tissue. Additionally due to the disaggregated nature of the cellular samples obtained by a brush biopsy, a slide prepared in connection with a brush biopsy will inevitably contain cells that overlap and overlay each other Consequently, an examining pathologist or specialized computer would want to build a constraint into the EDF algorithm such that pixels of objects can only come fiom one object or cell only and not from neighboring objects that are situated either on top or beneath the cell being imaged. For the purpose of this document this requirement is referred to as "PCO" or Pixel Constrained to an Object" Furthermore, there are objects of no diagnostic significance, such as cell walls or artifacts that should be deemphasized or eliminated fiom a composite image. SUMMARY OF THE INVENTION 5 In a first broad aspect the invention provides a method for creating an image of at least a portion of a three-dimensional semi-transparent biological specimen, said method comprising the steps of: a) positioning a microscope slide under a microscope, said microscope slide 5 comprising a three-dimensional biological specimen, said three-dimensional biological specimen comprising cells, said cells being located on different z-axis planes; b) obtaining a stack of images, said stack comprising a plurality of slices of said portion of said specimen, each of said plurality of slices comprising an image of said specimen obtained from a different focal distance, each of said slices comprising a io plurality of pixels, said stack of images comprising images of at least two cells each of said at least two cells located on different z-axis planes; c) performing at least one of a closing and a dilation on said plurality of slices; d) calculating the sharpness values of said plurality of pixels in said plurality of slices in said stack and storing max-sharpness values for said plurality of pixels; and is e) comparing the sharpness values for pixels in specific locations on a slice to stored max-sharpness values for pixels corresponding to said specific locations on previous slices in a stack. In a second broad aspect the invention provides a system for creating a digital 20 image of a semi-transparent three-dimensional biological specimen comprising: a semi-transparent cellular sample comprising disaggregated tissue, said cellular sample comprising cells overlying other cells; a computer apparatus; a microscope for obtaining a stack of image slices, each of said slices 25 corresponding to different focal planes; a data storage device for storing said images; software configured to perform a plurality of erosions to de-enhance a cell boundary, a number of said plurality of erosions being proportional to a pixel thickness of a cell boundary, calculate maximum sharpness values for pixels of said slices, perform at 30 least one of a closing and a dilation on said slices, and compare the maximum sharpness for pixels at various locations in a slice with that of pixels of corresponding locations from slices whose maximum sharpness have previously been calculated. 5A 4364974_1 (GHMatIers) P83561.AU 28105/2013 In a third broad aspect the invention provides a method of constructing a digital image of a three-dimensional biological specimen, said specimen comprising cells having cell walls and cell nuclei, said method comprising the steps of: 5 obtaining a three-dimensional biological specimen, said specimen comprising cells, said cells being located on different z-axis planes; acquiring a plurality of image slices of said specimen from a plurality of focal distances; de-enhancing images cell boundaries; 10 enhancing images of cell nuclei; selecting the sharpest pixels for each pixel location from among the plurality of slices; and copying said sharpest pixels into a composite image. 5B 43e4974_1 (GHMaters) P83581 AU 28/05/2013 WO 2009/026258 PCT/US2008/073526 An EDF system is contemplated which includes the ability of an EDF algorithm to (a) favor objects of interest and deemphasize objects of non-interest (b) wherein pixels ate constrained to come from that single object of interest only. Instead of'blindly capturing images that appear sharp, the EDF algorithm, which is the subject of the current invention, is adapted to select in-focus images or objects that are of diagnostic importance to a pathologist. Objects of no diagnostic importance are removed by performing a morphological erosion on all sharp appearing objects. In a preferred embodiment, an erosion algorithm is implemented which erodes pixels fiom each side of an image. Consequently, thin cell walls that appear sharp and in focus, but confer no diagnostically important infbimation, will be eliminated or significantly diminished The larger image of the cell's nucleus on the other hand, will survive the erosion process, thereby preserving the diagnostic objects When analyzing cellular specimens of the gastrointestinal tract, it is critical for an examining pathologist to observe the presence or absence of goblet cells. Goblet cells function as secretory cells and contain a sac of mucus in their interior. One ofthe key morphological char acteristics of goblet cells is a large central blob - that is light in contrast to the cell walls and the cell nucleus. Diagnostically important goblet cells will as such be preserved in the erosion step. To meet the PCO requirement, a morphological closing is applied A closing is a dilation followed by erosion, to close any small holes in the nuclei and goblet cells, in effect assuring that pixels come fPom the same object. In an embodiment of the current invention, a computer is used to analyze columnar epithelial cells taken from the gastrointestinal tract An examining pathologist prefers seeing diagnostic information over non-diagnostic ones The diagnostic ones include the goblet cells and nuclei of the columnar cells that are usually located within the tissue. The non diagnostic structures are the cell boundaries that appear readily on the 6 tissue surface which look like fish scales and often confuse the prior art EDF algorithms due to its greater sharpness value over columnar nuclei and goblet cells. 5 BRIEF DESCRIPRION OF THE DRAWING In order that the invention may be more clearly ascertained, embodiments will now be described, by way of example, with reference to the accompanying drawing, in which: Figure 1 is a block diagram showing the steps of prior art EDF algorithms. Figure 2 is a block diagram showing the steps of an EDF algorithm according to an 10 embodiment of the invention. Figure 3 shows two cells from three different z-distances. Figure 4 shows a composite image of the two cells shown in Fig. 2 both of which are in focus. 15 DETAILED DESCRIPTION OF THE INVENTION Embodiments of the present invention will now be described with reference to the above-identified figures of the Drawings. However, the Drawings and the description herein of the invention are not intended to limit the scope of the invention. It will be 20 understood that various modifications of the present description of the invention are possible without departing from the spirit of the invention. Also, features described herein may be omitted, additional features may be included, and/or features described herein may be combined in a manner different from the specific combinations recited herein, all without departing from the spirit of the invention. 25 Limited depth-of-focus is a deficiency in microscope imaging of thick, semi transparent biological samples that is effectively be rectified using EDF processing techniques that is the subject of the current invention. 30 Typically, in the preparation of a cellular specimen for pathology, a clinician will transfer and affix cells and/or tissue to a glass microscope slide. The slide is then sent to a laboratory for further processing and medical diagnosis. Such further processing 7 4364974_1 (GHMatters) P83581 AU 28/0512013 WO 2009/026258 PCT/US2008/073526 commonly includes staining the slide, for example with Feulgen or Papanicolaou stains, and applying a cover slip. The slide is then analyzed by either a pathologist or by a computer - or by some combination thereof In the current invention, however, a slide that comprises thiee-dimensionality undergoes still additional processing prior to being examined by either a pathologist and/or computer. Specifically, a captured digital image of the cellular specimen is further processed by the EDF system that is the subject of this invention which produces a novel enhanced image that presents diagnostically important cellular features that are in-focus everywhere. The software for performing the steps ofthe inventive system described herein are executed by a computer - as is well know in the art. The software may be implemented on programmed general purpose computer, a special purpose computer and/ox a microprocessor or the like, Fig. 1 shows various steps involved with prior art EDF algorithm At initialization, two temporary image buffers are cleared, the Z-index image buffer (step SI) and the max-sharpness image buffet (step S2). The max-sharpness image buffer represents the max-sharpness value for each of the pixel locations and the z-index buffer stores the image slice id number fiom where the max-sharpness pixel is located. In an initial step S4, a sobel or laplacian or any kind of sharpness operator is applied to calculate the sharpness value for each of the pixels in a z-image slice 20 The max sharpness values and slice id number are respectively stored in the max-sharpness and z index buffers. In step S7, for each z-index slice that is processed, the computer compares pixel values at each location of the slice with the stored max-sharpness for that pixel location. Note that for the first slice that is processed, the pixels are not compared to those of another slice as there is no preceding slice If a pixel's sharpness value is greater than the stored max-sharpness value in the corresponding pixel location of the max sharpness buffer, the software is directed to perform several actions, (1) replace the stored max-sharpness value in the max sharpness buffer with the new sharpness value at the corresponding pixel location S8, (2) replace the z-index value of the z-index buffer with the image slice id number at the corresponding image location S9. This process is 8 WO 2009/026258 PCT/US2008/073526 repeated until all z-image slices 20 have been processed. Once this is completed, the software is programmed to smooth the z-index buffer S10 using, for example, a Gaussian or Mean Filter to remove any incidental noise caused by the sharpness operator mentioned above. Using the smoothed z-index buffer, the software is configured to copy the sharpest pixels from each z-image slice S 11 to build the in-focus composite EDF image S12 Fig. 2 shows various steps through which an embodiment of this invention may be implemented. Preferably, at least two separate modules are injected into the prior art EDF algorithm - the Pre and Post processing module. Note that pre and post processing herein refers respectively to the step prior to and following the calculation of sharpness of each pixel in a slice The purpose of'these modules is to analyze the image and (a) to look for undesirable or non-diagnostic features and de-enhance those features, (b) look for preferable or diagnostic features and enhance those features, and (c) retain pixels from the same object to meet the PCO requirement. In an embodiment of the invention, a stack of successive images is acquired using a microscope In a preferred embodiment the focal distance between two slices is equal to half of the depth of field foi the microscope objective used to meet the Nyquist criteria of optimum sampling. Persons skilled in the art of optics can calculate this depth of field based on the numerical aperture of the objective and wavelength of light used However, one could increase the sampling distance provided one could afford the image quality degradation to gain additional processing speed. The Pre-Processing module is to enhance and de-enhance features that otherwise cannot be done after calculating the sharpness of the z-image slice In an embodiment of the invention, the Pre-Processing module comprises the step of performing an erosion S3, to remove thin edges of cell walls This erosion will have no effect on cell nuclei and goblets cells due to its blob size. The ultimate effect of this erosion process is to retain the sharpness features for the cell nuclei and goblet cells and deemphasize the sharpness of cell walls that have no diagnostic value. The number of erosion iterations required is 9 WO 2009/026258 PCT/US2008/073526 proportional to the pixel thickness of the cell boundary. Erosion algorithms, which are well known in the art, strip away a layer or layers of pixels from both the inside and the outside surfaces of an image. This is achieved by extending a lighter background into a darker image that is being eroded. For example, an image that is shaped like a doughnut will lose pixels forn the outside edge, and also fiorn the inside surface of the doughnut hole. However, an image that resembles a solid ball will be eroded from the outside surface only, Cell walls, which have an outside surface and an inside surface, will be eroded fPom both the inside and outside. The cell's nucleus, on the other hand, will only be eroded from the outside surface. The images of cell walls, which are very thin, will typically not survive or will be significantly diminished as a result of being eroded fPom two sides. Whereas the images of cell nuclei, which are much la ger and are only eroded flom the outside surface, will be only minimally affected by the erosion process As a result, the diagnostically important information represented by the cell nucleus will be preserved, while the unimportant images of cell walls will be eliminated or diminished. Subsequent to performing the erosion step S3 the software is programmed to calculate the sharpness value of each pixel in the z index slice S4. It will be understood by those of ordinary skill in the art, that the sharpness of an image can be calculated in any of various ways or in any combination thereof. F or example sharpness can be calculated (a) based on point process basis where sharpness is defined by comparing one pixel against z neighboring pixels only, (b) sharpness based on area process basis where sharpness is defined by comparing one pixels against it's x-y neighboring pixels (i.e laplacian, sobel and mutual information), and (c) sharpness based on the frequency space of the transformed image (i e fourier and wavelet transforms). Additionally, the image that is used for calculating sharpness can be RGB, grayscale, RGB converted to grayscale, or a weighted average of colors. One could also find the sharpness in each color separately and then combine them 10 WO 2009/026258 PCT/US2008/073526 After calculating the sharpness values, the software is configured to perform additional operations (the Post-Piocessing module) prior to advancing to the step of comparing pixel values (S7). In a prefered embodiment, as shown in Fig 2, the Post Processing module comprises the steps of performing two grayscale morphological operations The first operation comprises a grayscale closing S5 on the sharpness values This is needed to close any small holes within the cell nuclei and cell goblets, to assure all pixels of the object or cell are preserved to meet the PCO requirement The closing size is proportional to the blob size of the nuclei and goblet cell. The second operation comprises a (grayscale) dilation S6 to expand the sharpness value by one or more pixels wide to maintain the max sharpness or contrast between the nuclei or goblet boundary and the background In further embodiments of the invention any of several methods are employed for enhancing and de-enhancing specific objects or features of interest These include, but are not limited to. color, hue, object size and shape, aproi known x-y and z-location of objects to enhance/de-enhance objects. Having executed the Post Processing steps, the software is configured to compare the sharpness value of each pixel in a slice with those fiom slices processed prior thereto. Note, that for the first slice in a stack of z index slices this step (S7) is omitted - as there are no acquired pixel values with which to compare. Instead the step (S7) of comparing pixels in each location with those corresponding to that location on previous slices is performed beginning with the second slice in a stack and on any slice obtained thereafter The remaining steps are executed as described above (with reference to Fig- 1), whereby the sharpest pixels flom each of the slices are copied into one composite image S 11 resulting in a unique z-index slice that is in focus everywhere and which displays diagnostically important information - substantially to the exclusion of non-impoitant information. The z-index composite image is then stored on a data storage device, such as computer hardrive - for later access. 11 WO 2009/026258 PCT/US2008/073526 It should be noted that in other embodiments of the invention, the steps of performing an erosion (or sever al erosion) S3, a morphological closing S5 and a dilation S6 may be executed prior to the step of calculating the sharpness of the pixels within a given slice Moreover, in some embodiments, one or both of steps S5 and S6 may be omitted entirely. The smoothing filter is used to suppress any noise outside the nuclei and goblets cells that might have caused by the sharpening filter. It will be understood by those of ordinary skill in the art, that many types of operators meet the smoothing filter, including but not limited to Mean Filters, Grayscale Morphology, and Recursive Gaussian Filters. Ihe minimum size of the filter is proportional to the size of nuclei and goblet cells In one embodiment, it is not required to process the images at full resolution in which the image was captured To gain speed in processing in this invention, the software could be configured to downscale the z-images slices first prior to EDP processing Just before copying the sharpest pixels one would restore the smoothed z index image to the original resolution Additionally, or alternatively, in step S7, the system could be configured to compare only most, or some of the sharpest pixels with those of previous slices in the stack. In one preferred embodiment, the software optimizes the number of z-slices processed, adapting the number of slices to the depth of the sample, During motion of the z-stage, the software detects when all pixels of the image are out-of-fbcus, or have past the slice of maximum sharpness. In this case it is not necessary to take any more z-slices, thereby speeding up the processing. Once all the areas of interest on slide have been processed by the system and methods described herein above, the system is configured to stitch together the stored composite images in order to build a digital image of the entire specimen - or of areas of interest therefrom. Such stitching operations are well know in the art. After assembling the various composite slices - the resulting digital image of the specimen is in-focus 12 WO 2009/026258 PCT/US2008/073526 substantially everywhere and comprises substantially only diagnostically important cellular features. The inventive digital image may be displayed on a monitor fbi analysis by a pathologist, or may be analyzed by a trained computer Preferably, however, the specimen is analyzed initially by a computer and subsequently by a pathologist or cytopathologist. Specifically, in one embodiment of the invention, computerized analysis is conducted on the inventive image using the computer system disclosed in U.S patent No 's 6,297,044 and 6,284,482. In connection with system of the aforementioned U S patents, a computer using an image recognition system as a primary classifier yields a subset of cells that may be abnormal. A neural network system is employed as a secondary classifier to determine which of the subset of cells have the highest probability of being cancerous or pre-cancerous. A diagnosis is then made by a tertiary classifier - a trained pathologist - who examines the most abnormal appearing cells selected by the computer Alternatively a computer system may comprise only a primary classifier that selects the most abnormal appealing cells in a specimen Cells may be selected as abnormal based on morphology, abnormal keratinization or a combination of both - as compared to a stored cell library. In some embodiments, the computer works off of the assumption that there will be certain cells in a specimen that appear to be more abnormal than others The computer will thus rank cells according to their abnormal appearance. It will be understood by those skilled in the art that any of various specialized computer analytical systems may be used to analyze the inventive image In other embodiments of the invention, the inventive image is sent via intranet or Internet to a remote user or to a remote computer for analysis by either a remote user and/or for remote computer analysis For the purpose of this document a "slice" (or "z index slice") refers to a digital image of an area of a slide taken fiom a particular focal distance A "composite slice" refers to a digital image of an area of'a specimen that was processed by the inventive system - for example, as set forth in Fig. 2 Images may be RGB, grey scale or acquired 13 WO 2009/026258 PCT/US2008/073526 with a microscope filter wheel. It will be understood by those of ordinary skill in the art, that images may be acquired by any of various similar means The invention can be broadly defined as a method of constructing a digital image of at least a portion of a three-dimensional biological specimen, where e the specimen presents cells having cell walls and cell nuclei, comprising the steps of: acquiring a plurality of image slices of at least a portion of the specimen from a plurality of focal distances; de-enhancing images cell walls; enhancing images of cell nuclei; selecting the sharpest pixels for each pixel location from among the plurality of slices; and copying said sharpest pixels into a composite image. In one embodiment of the invention, the processing steps that are the subject of the invention are implemented in an on-line mode. In this embodiment, instead of applying some are all of the processing steps disclosed herein on the whole stack of input images, the output is updated one image at a time without saving the original images. In another embodiment, the capturing of images and processing of images are done in two passes In a first pass, images ate first acquired and stored. In a second pass, the stored images are processed by the processing steps of this invention. Referring to Fig. 3 shows two representative cells 24, 26 on a slide from three different z distances As shown, using a conventional microscope at high resolution, when cell 24 appears in-focus, cell 26 may be out of focus - and vice versa (slices A, B). At another focal distance (slice C) both cells are out of focus. If cells 24 and 26 were on the same plane - one could easily scan the entire slide at high resolution. However, when analyzing a specimen of the current invention, the cells are located at different levels (not shown) - making conventional scanning less than optimal. Referring to Fig. 4, in the inventive EDF system, cells 24 and 26 will appear in focus. 14 Having described the invention with regard to specific embodiments, it is to be understood that the description is not meant as a limitation since further modifications and variations may be apparent or may suggest themselves to those skilled in the art. It is intended that the present application cover all such modifications and variations as fall within the scope of the appended claims. It is to be understood that, if any prior art is referred to herein, such reference does not constitute an admission that such prior art forms a part of the common general knowledge in the art, in Australia or any other country. In the claims that follow and in the preceding description of the invention, except where the context requires otherwise due to express language or necessary implication, the word "comprise" or variations such as "comprises" or "comprising" is used in an inclusive sense, i.e. to specify the presence of the stated features but not to preclude the presence or addition of further features in various embodiments of the invention. 15 43649741 (GHMatters) P83581.AU 28/05/2013